Research On Image Stippling Stylized Algorithm

With the development of computer technology,non-photorealistic rendering has gradually become an important branch in the field of computer graphics research.It does not aim at pursuing image realism,and aims to reflect the artistry and aesthetics of images.As the main research direction,stylization can generate art works with some hand-painted styles,which are widely used in computer animation,film design and science education,and have high research value.As one of the objects of image stylization,stippling is a classic art technique that simulates the textures and light and dark changes of tones in images in the form of dots,expressing its artistic qualities.The automatic generation of stippling from images is the main research content of this thesis.In this thesis,an image-based real-time stippling generation algorithm is proposed.The stippling results obtained by this method are superior to the existing stippling generation algorithm in visual effects and numerical errors(PSNR,SSIM),and can achieve real-time rendering speed performance.The specific work of this thesis includes:1)Applying incremental Voronoi sampling sequences to achieve traditional stippling and diversified stippling.The incremental Voronoi sequence is a sampling algorithm.The sampling points are generated one by one,which can be increased indefinitely.If the sequence is stopped at any time,the distribution of sampling points with blue noise characteristics can be obtained,that is,the spatial distribution of these points is uniform and random.At the same time,it has great spectral characteristics.Therefore,the sequence can be used as a sampling point for generating stippling,and high-quality stippling results can be obtained.Traditional point painting is usually to draw fixed-size black dots on a white background,and the distribution of these points conveys the content and features of the input image.In this thesis,the drawing mode is realized,and at the same time,the drawing method of dot size and tone is realized,the quality of the stippling result is improved,and the result of the stippling is naturally transitioned in details and tone,and has higher visual effect.2)Apply the structural feature information and mask information of the input image to guide the drawing of the stippling.In the stippling algorithm of this thesis,In addition to the number of sampling points used in incremental Voronoi sequence and the gray value at the position of sampling points in the input image,the structural feature information and mask information of the input image can constrain the size of the dots.Therefore,the input image structure feature information and mask information are added in the traditional stippling and diversified stippling,which changes the distribution of the dots and obtains high-quality stippling results.This thesis compares and analyzes the stippling results obtained by each stippling mode from the visual effect and numerical error,and gradually adjusts the density,size and color of the sampling points,enhances the expressive power of the stippling,and is consistent with human visual perception.